EUROGRAPHICS 2013/ H.- C. Hege, A.

Vilanova

Medical Prize

O CTAV IS: A Virtual Reality System for

Clinical Studies and Rehabilitation
Eduard Zell1 , Eugen Dyck1 , Agnes Kohsik2 , Philip Grewe2 , David Flentge2 , York Winter3 , Martina Piefke4 , Mario Botsch1
1 Computer

Graphics & Geometry Processing, Bielefeld University

3 Cognitive Neurobiology, Humboldt University Berlin

2 Physiological
4 Neurobiology

Psychology, Bielefeld University

& Genetics of Behavior, Witten Herdecke University

surveillance camera

touch screen

joystick
pulse sensor
skin conductance
sensors
rotary encoder
collector ring

Figure 1: Two photographs and a simplified illustration of our O CTAV IS virtual reality (VR) system. Eight screens, arranged in
an octagon, provide a 360 panorama visualization of the virtual environment. Two door segments can be opened. Navigation
in the VR is performed through a modified office chair. Its orientation determines the movement direction. A throttle joystick
in the armrest controls the movement speed. Easy and natural interaction with objects is enabled through a simple touch screen
interface. Biosensors and surveillance cameras permit permanent patient observation by clinical staff.
Abstract
Brain function disorders, resulting for instance from stroke, epilepsy, or other incidents can be partially recovered
by rehabilitation training. Performing neuro-rehabilitation in virtual reality systems allows for training scenarios
close to daily tasks, is easily adaptable to the patients needs, is fully controllable by clinical staff, and guarantees
patient safety at all times. In this paper, we describe the O CTAV IS system, a novel virtual reality platform developed primary for clinical studies with and rehabilitation training of patients with brain function disorders. To meet
the special requirements for clinical use, our system has been designed with ease of use, ease of maintenance, patient safety, space and cost efficiency in mind. Our system has been successfully deployed to four hospitals, where
it is used for rehabilitation training and clinical studies. We report first results of these studies, demonstrating that
our system is immersive, easy to use, and supportive for rehabilitation purposes.

1. Introduction
Every year about 270.000 people in Germany suffer stroke.
Half of them remain disabled, which makes stroke the most
frequent reason for becoming disabled as an adult. Apart
from stroke, brain function disorders can also result from
cerebral traumata caused by accidents, as well as from psychiatric or neurological diseases (e.g., epilepsy).
c The Eurographics Association 2013.

Neuro-rehabilitation training can help to (at least partially) recover the lost cognitive abilities. Unfortunately it is
well known that the improvement gained in standard paperand-pencil tests cannot sufficiently be transferred to reallife scenarios. This is mainly because these standard tests
(1) train certain cognitive functions in isolation and (2) are
rather abstract and far from the problems in daily routine.

Zell et al. / O CTAV IS: A Virtual Reality System for Clinical Studies and Rehabilitation

VR technology helps to design more realistic training scenarios in highly immersive setups (see, e.g., [RBR05]). In
addition, VR training has the benefit that it can easily be adjusted to the specific needs or capabilities of the patient and
can perfectly be controlled by clinical staff.
In this paper, we describe the VR system O CTAV IS ,
which was developed during the interdisciplinary ERDF
project CITmed: Cognitive Interaction Technology for Medical Applications . Its main purpose is the diagnosis and
rehabilitation of the above mentioned brain function disorders. In particular, it has been designed to train memory,
spatial orientation and navigation, as well as higher order
executive functions like path planing. The first VR scenario
we have chosen for training these cognitive abilities in daily
tasks is grocery shopping in a virtual supermarket: Patients
have to memorize a list of shopping items, have to navigate
through the supermarket in order to find and buy each item,
and should improve their path through the supermarket over
multiple training sessions.
Currently, the O CTAV IS system is being evaluated in four
hospitals, where it achieves high acceptance by both staff
and patients and was shown to be supportive for rehabilitation purposes. The high acceptance rates by experts and nonexperts, old and young, and disabled and non-disabled people reflects well our effort to develop the O CTAV IS system
as a general VR training platform.
2. The O CTAV IS System
A VR system for rehabilitation training has to satisfy several (partially conflicting) requirements. For example, the
VR should on the one hand be highly immersive, but on the
other hand has to be reasonably cheap and with a small spatial footprint. The navigation should be intuitive and natural,
but also feasible for handicapped people. Therefore, we followed the advice of Bowman and McMahan [BM07], who
argue that rather than trying to increase each parameter responsible for immersion, it is more important to concentrate
on the parameters mostly involved for the task.
2.1. Hardware Setup
Although there exists a variety of VR systems, most of them
are not suitable for our purposes. VR training on standard
desktop PCs does not offer a sufficient degree of immersion.
CAVE or MiniCAVE systems are highly immersive, but far
too complex, space consuming, and costly to be used in a
clinical environment. Head-mounted displays (HMDs) lack
of self-perception in the virtual environment, which makes
interaction (e.g., buying items) less intuitive.

European Regional Development Fund, HighTech.NRW

http://citmed.uni-bielefeld.de

Our O CTAV IS system (cf. Figure 1) consists of eight 26"

LCDs that are arranged in a circle (an octagon) around
the patient, who is sitting in the center on a rotating office chair. The eight screens therefore provide an immersive 360 horizontal view of the virtual environment. The
screens are mounted on aluminum segments, two of which
are assembled as doors and provide an easy and safe entrance and exit for patients. In contrast to many multi-screen
VR systems that build on a distributed rendering solution
with one render client per view, our O CTAV IS is driven by
a single PC with three consumer-level graphics cards. Each
card provides three display ports, such that in addition to
the eight O CTAV IS screens one external operator display
can be connected. Our custom-tailored rendering framework
(Section 2.2) guarantees sufficient rendering performance by
fully exploiting CPU and GPU parallelism.
For the navigation in VR natural metaphors like realwalking, walking in place, or within a rotating sphere are
the most immersive, but are not suitable for disabled people. Navigation devices like game pads, keyboard and mouse
are not suitable for novice VR users, because they introduce
an additional abstraction layer. In our O CTAV IS system, we
have chosen a metaphor similar to an electronic wheelchair:
The movement direction is intuitively controlled by rotating the chair into the desired walking direction. Movement
speed (forward/backward) is controlled through a throttle
joystick in the armrest. Easy and intuitive interaction with
objects in the VR is enabled through touch screen displays.
This design choice significantly simplifies user interaction,
but comes at the price of bigger frames around the displays.
However, this was shown not to influence performance in
VR systems in [MPS11], which we could confirm in our
studies [DSPB12].
To account for medical requirements, some additional devices have been integrated (Figure 1, right). An optional
footrest allows even patients with hemiparesis to operate our
system. Biosensors attached to the fingers track the heart rate
and the skin conductance of the patient. Two surveillance
cameras give the operator a detailed overview about the patients action. Finally, a galvanic separation is incorporated
in order to secure patients from potential electric shocks. After fulfilling these special medical requirements our system
has been successfully CE-certified as a Class 1 medical device in Germany.
The resulting system, described in detail in [DZK 12],
is reasonably cheap (< 20k Euro), spatially compact (diameter < 120cm), easy to maintain (just one Windows PC),
andmost importantlyeasy to operate for patients: In our
studies all participants (elderly people, no VR experience,
healthy or stroke/epilepsy patients) succeeded in the virtual
shopping experiment, whereas in the CAVE-based study of
Renner et al. [RDS 10] most (young, healthy) novice users
failed to perform a very similar task in a virtual supermarket.
c The Eurographics Association 2013.

Zell et al. / O CTAV IS: A Virtual Reality System for Clinical Studies and Rehabilitation

2.2. Software Framework

Besides the hardware setup, the software framework is also
crucial for a successful VR system, in particular since our
rendering solution has to run on a single PC. Commercial
VR frameworks, such as, e.g., Virtools or Vizard, as well
as modified game engines, disqualify because of their price
and/or insufficient multi-GPU support. We therefore developed a slim, custom-tailored VR architecture, with a focus
on simple extensibility to new hardware devices, new experiment setups, and new virtual environments.
Our highly optimized 3D rendering pipeline fully exploits
the parallelism offered by our multi-core CPU and multiGPU workstation. The highly realistic supermarket model
used in our studies (Figure 1) consist of more than 4M triangles, which are rendered on eight screens at a rate of 70 fps,
which is more than 2 billion triangles per second. Note that
a frame-rate of at least 60 fps is crucial for avoiding cybersickness problems, in particular with non-VR-experts.
This is accomplished by a combination of low-level and
high-level optimizations for CPUs and GPUs: To feed the
eight screens, each GPU renders the scene up to three
times, distributed to a dedicated CPU thread per view. Lowlevel optimizations prevent GPUs from interfering with each
other, thereby guaranteeing the scalability of the whole system. All data is stored on the GPU and re-ordered for cache
efficiency. Higher-level optimization like geometry instancing and view-frustum culling further increase performance.
The whole rendering framework as well as the individual optimizations are described in more detail in [DSPB12].
3. Results of Clinical Studies
The O CTAV IS systems have been deployed to our collaborating medical partners about 1.5 years ago, where they are
evaluated since then. These institutions are: a stroke unit in a
clinic for neurology, an epilepsy center, a clinic for psychiatry and psychotherapy, and a neuro-rehabilitation clinic. The
feedback from the different hospitals attests that our system
is accepted and appreciated by staff and patients, not only
for its simplicity, but also for its suitability in their respective patient context.
First studies proved the O CTAV IS system to be immersive
and easy to use [DZK 12] as well as supportive for learning real-life cognitive abilities [GKF 12]. Here we want to
present results for training visuospatial cognition from two
different patient groups: First, 13 people with focal epilepsy,
being 1951 years old (mean=32.3, sd=10.0) and second, 11
stroke patients within the age range of 3476 (mean=61.0,
sd=15.2). In addition, we present results for 13 healthy senior people of age 6194 (mean=71.4, sd=10.8).
All groups performed the following training: On eight
consecutive days they had to buy a list of 20 items in a
grocery shopping task. On days 16 the same shopping list
c The Eurographics Association 2013.

(List A) was presented, memorized, and bought. On day 7 a

completely different distractive list (List B) was used. To test
how stable the learning of List A was, the participants had to
buy List A without presenting it again (free recall): once immediately after the distractive trial on day 7 and once more
on day 8. This training paradigm is based on the rationale of
classic neuropsychological tests of verbal learning and memory, such as the California Verbal Learning Test (CVLT) and
the Verbal Learning and Memory Test (VLMT).
Figure 2 shows the number of correctly bought items per
trial. In general, one can observe that the numbers increase
during days 16, and that the distractive trial on day 7 has
only a very small effect on the two subsequent trails with
free recall of List A. The negligible influence of the distractive list is in contrast to standard verbal learning tests and
proves the efficacy and stability of our multi-modal learning.
A repeated measures ANOVA proves statistical significance
of these improvements for seniors (p=0.003), stroke patients
(p=0.024) and epilepsy patients (p<0.001). Similarly to the
improvement in product score, the length of the walked trajectory and required time decreased during the eight days
(by about 30% and 40%, respectively), which shows an improvement of spatial orientation, map learning, and executive functions like path planning.
To analyze the improvement of general visuospatial performance, we compared the visuospatial abilities before and
after the training, using the comparable Rey Figure test (before) and Taylor Figure test (after). Concerning the visuoconstructive ability to assemble a figure from its components, all groups improved in average (seniors: 3.7%; stroke:
13.3%; epilepsy: 3.6%), but due to Wilcoxon calculus only
the patients showed statistical significance (stroke: z=-2.052,
p=0.040; epilepsy: z=-2.455, p=0.014). For the visuospatial
memory, participants had to remember and draw the figure
after 30 minutes. Again each group improved (seniors: 2.0%;
stroke: 18.8%; epilepsy: 15.7%), but only the patients did so
significantly (stroke: z=-1.956, p=0.050; epilepsy: z=-2.276,
p=0.023). It is plausible that the senior people did not improve significantly, since they had no brain injuries to recover from. These results clearly show that our system contributes to a general improvement in visuospatial cognition.
Finally, a questionnaire concerning the perceived efficacy of the training were performed after the day 8 (Figure 2, right). On a 06 scale (0=not al all, 3=average, 6=very
much), participants rated whether the O CTAV IS training was
interesting, motivating, and useful. All groups rate these
items very high, which is an important result, since it is
well known that internal motivation is crucial for treatment
success. Participants then rated whether they think they improved on memory, orientation, and their grocery shopping
performance. While both patient groups rate all learning areas above average, the senior people do so only for the memory gain. This is again plausible, since the healthy seniors did
not need help with orientation or shopping.

Untitled 1 Untitled 3 Untitled 5 Untitled 7 Untitled 9

productscore

Zell et al. / O CTAV IS: A Virtual Reality System for Clinical Studies and Rehabilitation
Correctly Bought Products

Efficacy Questionnaire

20
interesting
motivating

15

usefull
10
memory gain
5

Day 1 Day 2 Day 3 Day 4 Day 5 Day 6 Day 7 Day 7 Day 8

List A List A List A List A List A List A List B List A List A

Seniors

Stroke patients

orientation gain
improved shopping

Epilepsy patients

Figure 2: Results of O CTAV IS training for healthy seniors, stroke and epilepsy patients. Left: Number of correctly bought items
for each trial. Right: Ratings of training efficacy, concerning general appreciation (top) and perceived learning effects (bottom).
efficacy questions
interesting

4. Conclusion

motivating

From
usefullthe beginning the O CTAV IS system was designed and
developed
in a highly interdisciplinary effort by a team of
memory
computer scientists, psychologists, and medical scientists.
orientation
As such, and in contrast to most other VR platforms, it meets
orientation
the following crucial criteria for clinical use:

Acknowledgments
List A

List B

List A
free recall

This work was supported by the ERDF project CITmed:

Cognitive Interaction Technology for Medical Applications,
the DFG Centers of Excellence EXC 277 Cognitive Interaction Technology and EXC 257 NeuroCure.

real shoping

of
realEase
shoping

use: Our typical users are elderly patients, many of

them 0suffering
from
stroke,
and without
any PC or VR
1
2
3
4
5
6
experience. Still, thanks to its intuitive use even a 94 years
old participant
managed
to operate our
system right away.
senior
citizens
stroke
epilepsy

References
[BM07] B OWMAN D. A., M C M AHAN R. P.: Virtual reality:
How much immersion is enough? Computer 40, 7 (July 2007),
3643. 2

Maintenance: Since we employ a single PC, instead of a

rendering cluster, our system can be operated by clinical
staff without technical experience.

[DSPB12] DYCK E., S CHMIDT H., P IEFKE M., B OTSCH M.:

O CTAV IS: Optimization techniques for multi-GPU multi-view
rendering. Journal of Virtual Reality and Broadcasting 9, 6
(2012). 2, 3, 4

Cost efficiency: With costs of less than 20k Euro our system is cheaper than most multi-view VR systems and
therefore affordable for most neuro-rehabilitation clinics.

[DZK 12] DYCK E., Z ELL E., KOHSIK A., G REWE P., W IN TER Y., P IEFKE M., B OTSCH M.: O CTAV IS : An easy-to-use
VR-system for clinical studies. In Proceedings of Virtual Reality
Interaction and Physical Simulation (VRIPHYS) (2012), pp. 127
136. 2, 3

Space efficiency: With a diameter of about 120 cm the

O CTAV IS system fits easily into a typical hospital room.
Medical requirements: By using a robust chair with
footrest, usability and safety are given even for patients
with hemiparesis. Clinical staff can monitor the experiments and intervene at any moment. The system is a CEcertified Class I medical device in Germany.
Flexibility: While first clinical studies focused on the virtual supermarket, the system is easily extensible. More
specific studies (maze/city navigation, Morris water navigation) have been implemented and will be started soon.
These properties make our system a valuable platform
for multi-modal cognitive training in neuro-rehabilitation.
The first results of our clinical studies (shown in [DSPB12,
GKF 12] and this paper) confirm a high acceptance by
both medical staff and patients, and demonstrate our
O CTAV IS platform to be supportive in the rehabilitation of
cognitive disabilities resulting from stroke or epilepsy. Motivated by these results the consequent next step will be professional distribution of the O CTAV IS system.

[GKF 12] G REWE P., KOHSIK A., F LENTGE D., DYCK E.,
B IEN C., W INTER Y., B OTSCH M., M ARKOWITSCH H. J.,
P IEFKE M.: Learning real-life cognitive abilities in a novel
360 -virtual reality supermarket: A neuropsychological study of
healthy participants and patients with epilepsy. Journal of NeuroEngineering and Rehabilitation (2012). to appear. 3, 4
[MPS11] M C NAMARA A., PARKE F., S ANFORD M.: Evaluating performance in tiled displays: navigation and wayfinding. In
Proceedings of the 10th International Conference on Virtual Reality Continuum and Its Applications in Industry (2011), ACM,
pp. 483490. 2
[RBR05] ROSE F. D., B ROOKS B. M., R IZZO A. A.: Virtual
reality in brain damage rehabilitation: Review. CyberPsychology
and Behavior 8, 3 (2005), 241262. 2
[RDS 10] R ENNER P., DANKERT T., S CHNEIDER D., M ATTAR N., P FEIFFER T.: Navigating and selecting in the virtual supermarket: Review and update of classic interaction techniques. In Virtuelle und Erweiterte Realitt: 7. Workshop der
GI-Fachgruppe VR/AR (2010), pp. 7182. 2

c The Eurographics Association 2013.